The Agent Skills Directory

[COMMAND_EXECUTION] (MEDIUM): The skill is centered around executing system-wide configuration changes with root privileges using the Ansible become feature. While functional for its stated purpose, it provides a high-privilege execution environment for any commands contained within playbooks.
[COMMAND_EXECUTION] (MEDIUM): Documentation in references/ansible-runbook.md and logic in scripts/run-playbook.sh suggest or implement disabling SSH host key validation (ANSIBLE_HOST_KEY_CHECKING=False and -o StrictHostKeyChecking=accept-new). This practice bypasses a core SSH security control and makes the connection susceptible to Man-in-the-Middle (MITM) attacks.
[INDIRECT_PROMPT_INJECTION] (LOW): The skill processes local inventory files and playbooks which serve as the source of truth for operations. If these files are manipulated by an external actor, the agent may unknowingly execute unauthorized changes on the infrastructure.
Ingestion points: ansible/inventory/hosts.ini and playbook files in ansible/playbooks/.
Boundary markers: Absent; commands are executed based on the contents of the structured YAML/INI files.
Capability inventory: Execution of ansible and ansible-playbook with sudo permissions via scripts/run-playbook.sh.
Sanitization: None detected for the input variables used to construct the command line in scripts/run-playbook.sh.

ansible